The field of conformal prediction is moving towards developing more robust and efficient methods for generating prediction sets with guaranteed coverage. Researchers are exploring new approaches to improve the reliability and flexibility of conformal prediction, such as reducing the computational cost of robust conformal prediction and enabling the reuse of calibration sets. Notably, innovative methods are being proposed to achieve robustness with minimal computational overhead, and to adapt to changing distributions. Some papers are also investigating novel online learning settings, such as ambiguous online learning, and proposing new algorithms for multi-model online conformal prediction. Noteworthy papers include: One Sample is Enough to Make Conformal Prediction Robust, which proposes a single sample robust CP method that achieves state-of-the-art results with reduced computational cost. When Can We Reuse a Calibration Set for Multiple Conformal Predictions, which demonstrates how to reuse a calibration set with high probability while maintaining the desired coverage. Graph-Structured Feedback Multimodel Ensemble Online Conformal Prediction, which proposes a novel algorithm for selecting effective models and constructing smaller prediction sets.